Page 32 - Contributed Paper Session (CPS) - Volume 2

CPS1409 Rahma F.
known as a spatial lag variable, which is defined by attaching a spatial weight
matrix to the variable under study.
The spatial weight matrix is defined by considering the set of neighboring
locations of each location. The set of locations which are influential to
location in terms of the variables under study, are defined as the neighbors
of . Most of regional studies have used contiguity concept and geographical
distance to define the neighboring locations. Some definitions of the
neighbors can be found in Arbia (2006). The contiguity concept is based on
the administrative boundary, in which two locations that share common
administrative border are considered neighbors. There are two types of
neighbors based on distance, the critical cut off neighbors and nearest
neighbors. Both use any measure of distance between two locations. Physical
distance is generally represented by the Euclidian distance between the central
of two locations.
Due to the advance of transportation and communication technology,
economic interaction does not occur only between two adjacent locations.
While in terms of distance, it is not the flat physical distance which is
considered in the mobility of goods and labors. The distance in terms of the
transportation costs should be taken into account instead (Conley and Ligon
2002; Conley and Topa 2002). This concept of distance is defined as “economic
distance”. Conley and Ligon (2002) show that spatial interaction of the
economic growth among geographically neighboring countries is less intense
than the interaction among “economically” neighboring countries.
The weaknesses of the geographical – physical distance concepts motivate
this study to use other distance measure which includes the transportation
cost to capture the intensity of economic interaction between regencies/cities
in East Java. The “economic distance” is used to define the k nearest neighbors
for each 38 regencies/cities of East Java. The intensity of the spatial interaction
–spatial autocorrelation of economic productivity (GDP) will be tested based
on the defined distance. It compares the significance of the spatial
autocorrelation of the GDP among regencies/cities, when the neighbours are
defined by the physical distance and the economic distance. The better
distance measure should capture the spatial interaction better in terms of the
significance of the Moran I’s spatial autocorrelation test. Using the chosen
distance measure, this study also provides some analysis regarding the nature
of the spatial interaction of the economic productivity of East Java. It will be
useful for the policy maker to direct the growth in terms of interaction among
regencies/cities in this province.

2. Methodology
This study is based on regency/city as unit of observation and there are 38
regencies/cities in total. 2016 GDP of each regency/city (in thousand Rupiah)

21 | I S I W S C 2 0 1 9